CN111259590A - Construction method of concrete dam deformation safety monitoring model - Google Patents

Construction method of concrete dam deformation safety monitoring model Download PDF

Info

Publication number
CN111259590A
CN111259590A CN202010054570.4A CN202010054570A CN111259590A CN 111259590 A CN111259590 A CN 111259590A CN 202010054570 A CN202010054570 A CN 202010054570A CN 111259590 A CN111259590 A CN 111259590A
Authority
CN
China
Prior art keywords
model
dam
prediction
residual
value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202010054570.4A
Other languages
Chinese (zh)
Inventor
魏博文
罗绍杨
雷斌
李火坤
袁冬阳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanchang University
Original Assignee
Nanchang University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanchang University filed Critical Nanchang University
Priority to CN202010054570.4A priority Critical patent/CN111259590A/en
Publication of CN111259590A publication Critical patent/CN111259590A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • E—FIXED CONSTRUCTIONS
    • E02—HYDRAULIC ENGINEERING; FOUNDATIONS; SOIL SHIFTING
    • E02B—HYDRAULIC ENGINEERING
    • E02B1/00—Equipment or apparatus for, or methods of, general hydraulic engineering, e.g. protection of constructions against ice-strains
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/08—Computing arrangements based on specific mathematical models using chaos models or non-linear system models

Landscapes

  • Engineering & Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Civil Engineering (AREA)
  • Algebra (AREA)
  • Artificial Intelligence (AREA)
  • Computational Mathematics (AREA)
  • Structural Engineering (AREA)
  • Evolutionary Computation (AREA)
  • Nonlinear Science (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Mechanical Engineering (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a construction method of a concrete dam deformation safety monitoring model, which specifically comprises the steps of utilizing historical data of dam deformation observation, adopting a mixed frog leap algorithm (SFLA) with local optimization performance and global optimization performance to determine weight coefficients of each sub-model on the basis of establishing a stepwise regression model, utilizing an inversion analysis method of dam prototype data to determine physical and mechanical parameters of a dam, establishing a frog leap mixed model, and further obtaining the concrete dam deformation safety monitoring model. The method provided by the invention analyzes and predicts the residual error by using the chaos theory, and adds the residual error prediction item into the frog leap prediction hybrid model, thereby effectively solving the problems of good fitting effect and poor prediction result caused by the conventional dam displacement monitoring model without considering the influence of fitting residual error.

Description

Construction method of concrete dam deformation safety monitoring model
Technical Field
The invention relates to the technical field of dam operation safety monitoring and management, in particular to a construction method of a concrete dam deformation safety monitoring model.
Background
In recent years, concrete dams are developing towards high and large trends, the working condition and safety problems of the concrete dams are more and more emphasized by the dam industry, and the dams can cause great economic loss and mental loss to the country and the society once the dams are out of service. How to ensure the safe operation of the dam becomes an outstanding and important research subject in the field of dam safety monitoring. The deformation is one of important observed quantities for dam safety monitoring, and the structural damage and the working state of the dam structure under the action of internal and external environments can be intuitively and reliably reflected. How to establish a relatively accurate dam deformation prediction model according to dam prototype observation data has important significance for timely mastering the operation state of the dam and ensuring the safety of the dam.
The deterministic model simulates the effect quantity of the dam and the foundation under the action of actual load by using a finite element method, and is explained from the action relation mechanism of dam deformation and the effect quantity, but the calculation is complex when the temperature component is considered, and the operability is not high. The mixed model usually adopts a finite element method to calculate the water pressure component, and other components are selected according to the factors of the statistical model, thereby taking the advantages of the statistical model and the deterministic model into consideration. However, the models are all single prediction models, and the information mining method is single. The method is limited by external uncertainty factors, the current monitoring technology and analysis theory, the influence of main factors such as water pressure, temperature and aging is mainly considered, effective components contained in the chaotic residual sequence are ignored, and the precision of the deformation prediction model is limited. The existence of effective components in the residual sequence is effectively verified, and the research on the chaotic effect of the deformed residual sequence is still in an exploration stage. How to further improve the information mining scale of the prediction model and consider the effective components contained in the chaotic residual sequence has great significance for improving the prediction capability of the model.
The combined prediction model is widely applied to multiple fields because of the combination of the advantages of multiple models and the powerful information mining function, and the combined prediction model is utilized to establish a novel prediction model which not only considers the influence of measured value residual errors, but also can better map the nonlinear relation between effect quantity and influence factors, and has certain theoretical value and practical significance for improving the prediction precision. Inspired by the application of SFLA in the fields of project management, economic optimization configuration, medicine, electric power and other nonlinear optimization problems, an improved dam deformation hybrid prediction model based on SFLA is proposed in the research. The invention adopts a mixed frog-leap algorithm (SFLA) with local optimization performance and global optimization performance to determine the weight coefficient of each submodel, and utilizes an inversion analysis method of dam prototype data to determine the physical and mechanical parameters of the dam and establish a frog-leap mixed model. Meanwhile, aiming at the problems that the conventional dam displacement monitoring model has good fitting effect and poor prediction result due to the fact that the influence of fitting residual is not considered, the residual is analyzed and predicted by using a chaos theory, a residual prediction item is added into a leapfrog prediction hybrid model, and accordingly the leapfrog type hybrid prediction model considering the chaos characteristic of the concrete dam displacement residual is constructed so as to improve the prediction accuracy of the model.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides a method for constructing a concrete dam deformation safety monitoring model, which is realized by the following technical scheme:
a method for constructing a concrete dam deformation safety monitoring model comprises the following steps:
s1, constructing an improved concrete dam safety monitoring displacement mixed model based on SFLA;
s2, predicting chaotic residual errors of the fitting sequence;
s3, overlapping the residual prediction item and the mixed prediction model;
and S4, constructing a concrete dam deformation safety monitoring model.
In step S1, the SFLA-based improved concrete dam safety monitoring displacement hybrid model is constructed, and the construction method thereof is as follows:
s11, constructing a frog-leaping mixed prediction sub-model:
the dam in long-term service has great change in structural parameters or material parameters due to the accumulation of damage, and the operation state of the dam can be mastered by inverting the main physical and mechanical parameters of the dam body and the foundation;
finding out true water pressure component delta from analysis of prototype observation dataHThen assuming elastic modulus E of dam and bedrockc0、Er0Estimating the water pressure component δ 'by structural analysis'HWherein
Ec＝Ec0δ′H/δ H①
Er＝Er0δ′H/δ H②
Calculating a water pressure component by using a finite element method after the average elastic modulus of the dam body and the bedrock is obtained through inversion of the formula, and adding the water pressure component and the temperature component and the aging component under the statistical model to obtain a mixed model of dam deformation, namely a sub-model of the frog leap mixed prediction model;
s12, determining a sub-model weight coefficient by using a mixed frog-leaping algorithm SFLA;
and S13, constructing a displacement optimization hybrid prediction model.
In step S13, the displacement optimization hybrid prediction model is constructed by the following specific processes:
{δtt is 1,2, …, n, which is displacement monitoring data sequence in a certain time period of the dam, if there are p feasible modeling methods to establish a mixed model, it is set
Figure BDA0002372370490000021
The predicted value of the ith prediction model (sub-model) at the time t, the weight coefficient of the ith prediction model in the hybrid prediction (wherein,
Figure BDA0002372370490000022
) Then, the predicted value of the combined prediction model at time t can be expressed as:
Figure BDA0002372370490000031
in the formula, deltatAs model predictive value, liIn order to be the weight coefficient,
therefore, the prediction error of the hybrid prediction model at time t is:
Figure BDA0002372370490000032
accordingly, the sum of the square of the prediction errors of the hybrid prediction model at all the time can be obtained:
Figure BDA0002372370490000033
the hybrid prediction model with the criterion of least sum of squared prediction errors can be expressed as:
Figure BDA0002372370490000034
the chaotic residual prediction of the fitting sequence in the step S2 includes the following steps:
1) phase space reconstruction of residual sequence: for residual sequences ε1,ε2,…,εnIt can be reconstructed as an m-dimensional phase space:
Figure BDA0002372370490000035
in the formula, phi is a phase point; z is the phase point number, and Z is n- (m-1) tau; m is the embedding dimension; τ is the delay time;
2) determination of the embedding dimension m: determining an embedding dimension m, selecting a G-P algorithm which is typically proposed by Grassberger and Procaccia and used for calculating an attractor association dimension d from a time sequence, wherein the calculation process is as follows:
the correlation integral is first calculated:
Figure BDA0002372370490000036
wherein Z is the number of phase points; i phii-ΦjI represents the phase point phiiAnd phase point phijThe euclidean distance between; c (r) in phase spaceProbability that the distance between two phase points is less than r, wherein the phase points with the distance less than r are called correlation phase points; θ (x) is the Heaviside unit function:
Figure BDA0002372370490000041
when the embedding dimension m takes a different value, the attractor association dimension d satisfies a log-linear relationship with c (r):
Figure BDA0002372370490000042
as the embedding dimension m increases, dmThe value is also increased, when m reaches a certain value, m and d are continuously increasedmThe value will remain unchanged, i.e. reach a stable value dmThe m value is the embedding dimension;
3) determination of the delay time τ:
the formula is calculated by adopting a depolarization complex autocorrelation method:
Figure BDA0002372370490000043
in the formula, n is the length of a residual sequence; m is the embedding dimension; epsiloniIs the ith sample in the sequence;
Figure BDA0002372370490000044
is the average of the sequences. RεThe value of (d) is continuously reduced along with the increase of the value of tau when R isεWhen the value is reduced to 1-1/e of the initial value, the corresponding tau value is the delay time required by the reconstruction phase space;
4) chaotic characteristic identification and residual prediction of residual sequences:
the prediction of the chaotic time sequence is briefly explained by taking a Lyapunov exponent method as an example;
if necessary to the residual error epsiloni+1Predicting and selecting phase point phiiTo predict the center point, and assume the phase point ΦiHas a nearest neighbor of philWherein phiiAnd philIn betweenThe distance is as follows:
Figure BDA0002372370490000045
the expression of the maximum Lyapunov exponent prediction model is as follows:
Figure BDA0002372370490000046
in the above formula only phii+1Last component of (e)i+1Is unknown, so for εi+1Making predictions is feasible;
and (3) superposing the residual prediction item and a prediction model without considering the residual chaotic effect to obtain a dam displacement frog leaping mixed prediction model considering the chaotic residual:
Figure BDA0002372370490000047
in the formula, epsilon is a dam displacement residual error predicted value obtained by adopting a chaos theory.
The method has the beneficial effects that:
the method comprises the steps of determining the weight coefficient of each sub-model by adopting a mixed frog leap algorithm (SFLA) with local optimization performance and global optimization performance, determining the physical and mechanical parameters of a dam by utilizing an inversion analysis method of dam prototype data, establishing a frog leap mixed model, analyzing and predicting residual errors by using a chaos theory, adding a residual error prediction item into the frog leap prediction mixed model, solving the problems of good fitting effect and poor prediction result of a conventional dam displacement monitoring model due to the fact that the influence of fitting residual errors is not considered, and improving the prediction accuracy of the monitoring model.
Drawings
FIG. 1 is a flow chart of the construction of a concrete dam deformation safety monitoring model according to the present invention;
FIG. 2 is a view showing a vertical arrangement of a dam;
FIG. 3 is an upstream water level process line;
FIG. 4 is a 2# dam section finite element model;
FIG. 5 is a convergence process of the mixed frog-leap algorithm (SFLA);
FIG. 6 is a horizontal displacement fitting process line of a mixed frog-leap algorithm (SFLA) mixed model;
FIG. 7 is a residual sequence process line;
FIG. 8 is a graph of lnC (r) versus lnr;
FIG. 9 shows the prediction result of the horizontal displacement of PL5 measurement point;
FIG. 10 shows the prediction results of the horizontal displacement of PL2 measurement point.
Detailed Description
The invention is further described below with reference to the figures and examples.
Example (b): see fig. 1, fig. 2.
Fig. 1 is a flowchart of a method for constructing a concrete dam deformation safety monitoring model according to the present invention, and the embodiment specifically includes: the maximum dam height of a roller compacted concrete gravity dam is 113.0m, the total length is 308.5m, the dam crest elevation is 179.0m, the normal water storage level is 173.0m, and the regulated storage capacity is 11.22 hundred million m3Check flood level 177.8m, corresponding total reservoir capacity 20.35 hundred million m3. The dam body is not provided with longitudinal seams and is divided into 6 dam sections by 5 transverse seams. In order to ensure the safe operation of the dam and the underground powerhouse, more comprehensive monitoring items such as deformation, seepage, temperature, stress strain and the like are arranged on the surface (or inside) of a main building. The dam deformation monitoring items mainly comprise a Plumb Line (PL), a reverse plumb line (IP), a tension line, a sight line and the like, and leveling measurement is adopted for vertical displacement monitoring. The front vertical lines and the back vertical lines of the dam are used for monitoring the displacement of the dam crest and the water level inside the dam body in the river direction and the vertical water flow direction, an automatic monitoring system is connected in 10 months in 2002, and the arrangement scheme is shown in fig. 2.
Because the non-overflow dam section is less disturbed by water flow such as flood discharge and the like, and the monitoring data is regular, long-term monitoring data of horizontal displacement of a measuring point PL5(179m) at 1 month and 1 day of 2009-2014 12 months and 31 months of 2014 of the 2# non-overflow dam section are selected for analysis, and input quantities such as a water pressure component, a temperature component and an aging component are all subjected to zeroing treatment on the first monitoring day (namely 1 month and 1 day of 2009). The process line of the upstream water level, the process line of the rainfall amount in the dam site area and the process line of the temperature measurement value in the monitoring period are shown in fig. 3.
Figure BDA0002372370490000061
And (3) respectively establishing a statistical model for the measuring points PL 5:
(1) stepwise regression model
Dam displacement is mainly influenced by factors such as reservoir water level, temperature, ageing and the like, and the statistical model expression is as follows:
Figure BDA0002372370490000071
in the formula, ai、b1i、b2i、c1、c2Is a regression coefficient; H. h0The water depth at the upstream of the monitoring day and the initial monitoring day respectively; t is the cumulative number of days from the modeling series measurement day to the initial monitoring day, t0Accumulating the number of days from the first day of the modeling data series to the initial monitoring day; theta is the cumulative number of days from the monitoring day to the initial monitoring day multiplied by 0.01; theta0The cumulative number of days from the first measurement day to the beginning measurement day of the modeling data series was multiplied by 0.01.
(2) BP model
Selecting a three-layer BP neural network consisting of an input layer, an output layer and a hidden layer. The input layer is mainly the influence factors of dam displacement, such as water pressure, temperature, aging and the like, and the expressions of the three influence factors are respectively expressed by the formula (22), so that the total number of input layer nodes of the network is 9; the network output is the dam displacement, so the number of output layer nodes is 1; the number of hidden layer nodes is taken to be 11. The network training parameters are set as follows:
network learning rate net.trainParam.lr is 0.01;
the maximum training time of the network net, trainParam, epochs is 2000;
maximum allowed error of training net. trainparam. goal ═ 0.005.
Establishing a finite element model of the No. 2 dam section, and obtaining the elastic modulus E of the dam concrete through inversionc2.51GPa, elastic modulus E of dam foundationr1.47 GPa. Base in statistical modelOn the basis, a finite element method is used for calculating a water pressure component to replace a water pressure component in the statistical model, and the mixed model is obtained. On the basis, an SFLA (small form-factor pluggable) mixed prediction model (short for SFLA mixed model) and an SFLA mixed prediction model (short for chaotic SFLA mixed model) considering chaotic residual errors are respectively established by using the SFLA and the chaotic theory. The finite element model of the No. 2 dam segment is mainly composed of eight-node hexahedron isoparametric units and a small number of pentahedron and tetrahedron units, and comprises 51775 units and 47964 nodes in total, and the overall effect is shown in FIG. 4.
Taking observation data in 2009-2013 as a training sample, and determining the optimal weight coefficient of a sub-model in a combined model by using SFLA (small form-factor analysis) on the basis of taking two mixed models established based on stepwise regression and BP (back propagation) neural network as the sub-model, wherein the parameters of the leapfrog algorithm are set as follows: the number of the swarm grouping M is 50, the number N of the frogs in each group is 30, the number Ne of the iteration in the group is 30, and the total evolutionary number of the swarm is MAXGEN is 50. Fig. 5 is a SFLA convergence procedure. As can be seen from fig. 5, the frog-leap algorithm has good convergence, and after 3 iterations, the fitness value is substantially stable, and the weight coefficient l1 of the hybrid model established based on stepwise regression is 0.4657, and the weight coefficient l2 of the hybrid model established based on the BP neural network is 0.5728, so that the SFLA model can be established. Fig. 6 and 7 are the SFLA hybrid model horizontal displacement fit process line and the residual sequence process line, respectively.
And taking the residual error of the SFLA mixed model displacement fitting (see figure 7) as a residual error sequence to be analyzed, and marking as { epsilon ∈i}. Firstly, determining an embedding dimension m required by phase space reconstruction by adopting a G-P algorithm, wherein the value of the embedding dimension m cannot be too large or too small, if m is too small, an attractor of the phase space is selfed, and false adjacent points are easy to appear; if m is too large, the phase points are too far apart and the true neighbors are easily lost. A relationship graph of lnc (r) and lnr is obtained by G-P algorithm, taking m as 2,3, … and 8, respectively, as shown in fig. 8.
As can be seen from the figure, under the condition of different values of the embedding dimension m, straight line portions exist in the curves of lnc (r) and lnr, and the slope of the straight line portions is the attractor dimension corresponding to the different embedding dimensions m. It can be seen that when m is 6, the slope of the straight line portion is substantially constant, andapproximately parallel when m is 7, 8, i.e. when the attractor dimension reaches a stable value, calculated as 3.23, which is a non-integer, indicates the sequence of residuals analyzed, { ε }iHas chaotic characteristics.
After the embedding dimension m is determined, the delay time τ is determined by means of a deskew complex autocorrelation method. The delay time τ was calculated to be 2. Obtaining delay time and embedding dimension and comparing residual sequences { epsiloniAfter phase space reconstruction is carried out, the maximum Lyapunov index of the residual sequence is calculated to be lambda1When 0.2017 > 0, the residual sequence { epsiloniHas chaotic characteristics. Decision residual sequence [ epsilon ]iAnd after the chaos time sequence is subjected to phase space reconstruction, predicting the residual error by adopting a chaos theory. The maximum Lyapunov index prediction method is adopted in the text, and the specific calculation method is shown in formula (20). Get the pair residual error sequence { epsiloniPredicted value from 1 month and 1 day of 2014 to 31 months of 2014 (marked as { epsilon'i})。
Predicting residual error to be epsilon'iAnd (4) adding the chaos residual factor into the SFLA mixed model to form a dam displacement chaos mixed model (chaos SFLA mixed model) considering the chaos residual factor. In order to compare the prediction effects of the chaotic SFLA mixed model, the stepwise regression model, the BP neural network model and the mixed model based on the stepwise regression and the BP neural network, the displacement prediction results of the six models on the PL5 measuring point 2014 from 1 month 1 to 12 months 31 and are plotted in FIG. 9.
In order to test the prediction effect of the established model, the horizontal displacement long-term monitoring data of the PL2 measuring point is used as training data by using the PL2(140m) measuring point from 2009 to 2013 of the No. 2 dam section as a reference point and by using the method in the text, the displacement of the PL2(140m) measuring point from 2014 1 month to 2014 12 months and 31 days is also predicted, and the prediction result is drawn in fig. 10.
In order to further compare the prediction accuracy of the chaos SFLA mixed model with the SFLA mixed model, the stepwise regression mixed model, the BP mixed model, the stepwise regression model and the predicted value of the BP model, the statistical indexes of the prediction results of the six models with the PL5 measuring point and the PL2 measuring point, such as the average error ME, the mean square error MSE and the average relative error MAPE, are calculated respectively, and are shown in the table 1.
TABLE 1 comparison of statistical indices for six prediction models
Figure BDA0002372370490000091
As can be seen from Table 1, compared with the traditional model, the MAE, MSE and MAPE are closer to 0, namely closer to the perfect model, which shows that the prediction precision of the combined model is better than that of the conventional monitoring model, thereby verifying the rationality and scientificity of the invention.
Although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the spirit and scope of the invention.

Claims (4)

1. A method for constructing a concrete dam deformation safety monitoring model comprises the following steps:
s1, constructing an improved concrete dam safety monitoring displacement mixed model based on SFLA;
s2, predicting chaotic residual errors of the fitting sequence;
s3, overlapping the residual prediction item and the mixed prediction model;
and S4, constructing a concrete dam deformation safety monitoring model.
2. The method for constructing the concrete dam deformation safety monitoring model according to claim 1, wherein the step S1 is implemented by constructing an SFLA-based improved concrete dam safety monitoring displacement hybrid model, and the construction method comprises the following steps:
s11, constructing a frog-leaping mixed prediction sub-model:
the dam in long-term service has great change in structural parameters or material parameters due to the accumulation of damage, and the operation state of the dam can be mastered by inverting the main physical and mechanical parameters of the dam body and the foundation;
finding out true water pressure component delta from analysis of prototype observation dataHThen assuming elastic modulus E of dam and bedrockc0、Er0Estimating the water pressure component δ 'by structural analysis'HWherein
Ec＝Ec0δ′H/δH①
Er＝Er0δ′H/δH②
Calculating a water pressure component by using a finite element method after the average elastic modulus of the dam body and the bedrock is obtained through inversion of the formula, and adding the water pressure component and the temperature component and the aging component under the statistical model to obtain a mixed model of dam deformation, namely a sub-model of the frog leap mixed prediction model;
s12, determining a sub-model weight coefficient by using a mixed frog-leaping algorithm SFLA;
and S13, constructing a displacement optimization hybrid prediction model.
3. The method for constructing the model for monitoring the deformation safety of the concrete dam according to claim 2, wherein the displacement optimization hybrid prediction model is constructed in step S13 by the following specific processes:
{δtt is 1,2, …, n, which is displacement monitoring data sequence in a certain time period of the dam, if there are p feasible modeling methods to establish a mixed model, it is set
Figure FDA0002372370480000011
The predicted value of the ith prediction model (sub-model) at the time t, the weight coefficient of the ith prediction model in the hybrid prediction (wherein,
Figure FDA0002372370480000012
) Then, the predicted value of the combined prediction model at time t can be expressed as:
Figure FDA0002372370480000013
in the formula, deltatAs model predictive value, liIn order to be the weight coefficient,
therefore, the prediction error of the hybrid prediction model at time t is:
Figure FDA0002372370480000021
accordingly, the sum of the square of the prediction errors of the hybrid prediction model at all the time can be obtained:
Figure FDA0002372370480000022
the hybrid prediction model with the criterion of least sum of squared prediction errors can be expressed as:
Figure FDA0002372370480000023
4. the method for constructing a model for monitoring the deformation safety of a concrete dam according to claim 1, wherein the chaotic residual prediction of the fitting sequence in step S2 comprises the following steps:
1) phase space reconstruction of residual sequence: for residual sequences ε1,ε2,…,εnIt can be reconstructed as an m-dimensional phase space:
Figure FDA0002372370480000024
in the formula, phi is a phase point; z is the phase point number, and Z is n- (m-1) tau; m is the embedding dimension; τ is the delay time;
2) determination of the embedding dimension m: determining an embedding dimension m, selecting a G-P algorithm which is typically proposed by Grassberger and Procaccia and used for calculating an attractor association dimension d from a time sequence, wherein the calculation process is as follows:
the correlation integral is first calculated:
Figure FDA0002372370480000025
wherein Z is the number of phase points; i phii-ΦjI represents the phase point phiiAnd phase point phijThe euclidean distance between; c (r) represents the probability that the distance between two phase points in the phase space is less than r, and the phase points with the distance less than r are called correlation phase points; θ (x) is the Heaviside unit function:
Figure FDA0002372370480000026
when the embedding dimension m takes a different value, the attractor association dimension d satisfies a log-linear relationship with c (r):
Figure FDA0002372370480000027
as the embedding dimension m increases, dmThe value is also increased, when m reaches a certain value, m and d are continuously increasedmThe value will remain unchanged, i.e. reach a stable value dmThe m value is the embedding dimension;
3) determination of the delay time τ:
the formula is calculated by adopting a depolarization complex autocorrelation method:
Figure FDA0002372370480000031
in the formula, n is the length of a residual sequence; m is the embedding dimension; epsiloniIs the ith sample in the sequence;
Figure FDA0002372370480000032
is the average of the sequences. RεThe value of (d) is continuously reduced along with the increase of the value of tau when R isεWhen the value is reduced to 1-1/e of the initial value, the corresponding tau value is the delay time required by the reconstruction phase space;
4) chaotic characteristic identification and residual prediction of residual sequences:
the prediction of the chaotic time sequence is briefly explained by taking a Lyapunov exponent method as an example;
if necessary to the residual error epsiloni+1Predicting and selecting phase point phiiTo predict the center point, and assume the phase point ΦiHas a nearest neighbor of philWherein phiiAnd philThe distance between them is:
Figure FDA0002372370480000033
the expression of the maximum Lyapunov exponent prediction model is as follows:
Figure FDA0002372370480000034
in the above formula only phii+1Last component of (e)i+1Is unknown, so for εi+1Making predictions is feasible;
and (3) superposing the residual prediction item and a prediction model without considering the residual chaotic effect to obtain a dam displacement frog leaping mixed prediction model considering the chaotic residual:
Figure FDA0002372370480000035
in the formula, epsilon is a dam displacement residual error predicted value obtained by adopting a chaos theory.
CN202010054570.4A 2020-01-17 2020-01-17 Construction method of concrete dam deformation safety monitoring model Pending CN111259590A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010054570.4A CN111259590A (en) 2020-01-17 2020-01-17 Construction method of concrete dam deformation safety monitoring model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010054570.4A CN111259590A (en) 2020-01-17 2020-01-17 Construction method of concrete dam deformation safety monitoring model

Publications (1)

Publication Number Publication Date
CN111259590A true CN111259590A (en) 2020-06-09

Family

ID=70952265

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010054570.4A Pending CN111259590A (en) 2020-01-17 2020-01-17 Construction method of concrete dam deformation safety monitoring model

Country Status (1)

Country Link
CN (1) CN111259590A (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112100711A (en) * 2020-08-10 2020-12-18 南昌大学 ARIMA and PSO-ELM-based concrete dam deformation combined forecasting model construction method
CN112257973A (en) * 2020-09-04 2021-01-22 华电电力科学研究院有限公司 Hydropower station dam safety management effect integral dynamic calculation method based on dam safety monitoring result
CN112464570A (en) * 2020-12-10 2021-03-09 中国电建集团华东勘测设计研究院有限公司 BP-LSTM-based intelligent dam monitoring physical quantity prediction method
CN112507591A (en) * 2020-12-11 2021-03-16 四川大学 Gravity dam crest displacement monitoring method based on separation type hybrid modeling technology
CN112504934A (en) * 2020-11-23 2021-03-16 中国水利水电科学研究院 Concrete dam seepage pressure prediction and monitoring threshold determination method
CN113468795A (en) * 2021-04-02 2021-10-01 水利部南京水利水文自动化研究所 Universal building method for different parameter material partition pouring/filling dam deformation mixed model
CN113742939A (en) * 2021-09-15 2021-12-03 水利部南京水利水文自动化研究所 Construction method of concrete dam effect quantity temperature component model in operation period
CN114547951A (en) * 2022-04-24 2022-05-27 浙江远算科技有限公司 Dam state prediction method and system based on data assimilation
CN115758252A (en) * 2022-09-15 2023-03-07 南京河海南自水电自动化有限公司 Monitoring information real-time processing and analyzing method based on multi-source information fusion technology

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106370329A (en) * 2016-09-06 2017-02-01 河海大学 Roller compacted concrete dam structural condition monitoring system and monitoring method
CN108846199A (en) * 2018-06-12 2018-11-20 华能澜沧江水电股份有限公司 Extra-high arch dam deformation Time-space serial prediction technique based on Spatio-temporal modeling

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106370329A (en) * 2016-09-06 2017-02-01 河海大学 Roller compacted concrete dam structural condition monitoring system and monitoring method
CN108846199A (en) * 2018-06-12 2018-11-20 华能澜沧江水电股份有限公司 Extra-high arch dam deformation Time-space serial prediction technique based on Spatio-temporal modeling

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
熊威: "顾及多效应的混凝土坝位移联合预报与监控分析", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 *
王伟等: "混凝土坝变形组合预报模型的蛙跳建模方法", 《水利水运工程学报》 *
魏博文等: "融合混沌残差的大坝位移蛙跳式组合预报模型", 《武汉大学学报·信息科学版》 *

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112100711A (en) * 2020-08-10 2020-12-18 南昌大学 ARIMA and PSO-ELM-based concrete dam deformation combined forecasting model construction method
CN112257973A (en) * 2020-09-04 2021-01-22 华电电力科学研究院有限公司 Hydropower station dam safety management effect integral dynamic calculation method based on dam safety monitoring result
CN112257973B (en) * 2020-09-04 2023-11-28 华电电力科学研究院有限公司 Hydropower station dam safety management achievement point dynamic calculation method based on dam safety monitoring achievement
CN112504934A (en) * 2020-11-23 2021-03-16 中国水利水电科学研究院 Concrete dam seepage pressure prediction and monitoring threshold determination method
CN112504934B (en) * 2020-11-23 2021-09-21 中国水利水电科学研究院 Concrete dam seepage pressure prediction and monitoring threshold determination method
CN112464570A (en) * 2020-12-10 2021-03-09 中国电建集团华东勘测设计研究院有限公司 BP-LSTM-based intelligent dam monitoring physical quantity prediction method
CN112464570B (en) * 2020-12-10 2022-12-20 中国电建集团华东勘测设计研究院有限公司 BP-LSTM-based intelligent dam monitoring physical quantity prediction method
CN112507591A (en) * 2020-12-11 2021-03-16 四川大学 Gravity dam crest displacement monitoring method based on separation type hybrid modeling technology
CN112507591B (en) * 2020-12-11 2022-11-18 四川大学 Gravity dam crest displacement monitoring method based on separation type hybrid modeling technology
CN113468795B (en) * 2021-04-02 2023-10-31 水利部南京水利水文自动化研究所 General construction method for different parameter material partition casting/filling dam deformation mixed model
CN113468795A (en) * 2021-04-02 2021-10-01 水利部南京水利水文自动化研究所 Universal building method for different parameter material partition pouring/filling dam deformation mixed model
CN113742939A (en) * 2021-09-15 2021-12-03 水利部南京水利水文自动化研究所 Construction method of concrete dam effect quantity temperature component model in operation period
CN114547951A (en) * 2022-04-24 2022-05-27 浙江远算科技有限公司 Dam state prediction method and system based on data assimilation
CN114547951B (en) * 2022-04-24 2022-07-22 浙江远算科技有限公司 Dam state prediction method and system based on data assimilation
CN115758252A (en) * 2022-09-15 2023-03-07 南京河海南自水电自动化有限公司 Monitoring information real-time processing and analyzing method based on multi-source information fusion technology
CN115758252B (en) * 2022-09-15 2024-04-16 南京河海南自水电自动化有限公司 Monitoring information real-time processing and analyzing method based on multisource information fusion technology

Similar Documents

Publication Publication Date Title
CN111259590A (en) Construction method of concrete dam deformation safety monitoring model
Chen et al. Multi-kernel optimized relevance vector machine for probabilistic prediction of concrete dam displacement
Su et al. Time-varying identification model for dam behavior considering structural reinforcement
CN108549770B (en) Adaptive inversion method for rock-fill dam material parameters based on QGA-MMRVM
Li et al. Hydrostatic seasonal state model for monitoring data analysis of concrete dams
CN110807508B (en) Bus peak load prediction method considering complex weather influence
CN111428419A (en) Suspended sediment concentration prediction method and device, computer equipment and storage medium
Hellgren et al. Performance of data-based models for early detection of damage in concrete dams
CN112836789A (en) Ground connection wall deformation dynamic prediction method based on composite neural network algorithm
CN115659609A (en) DTW-DCRNN-based chemical industry park noise prediction method
Xue et al. PREDICTION OF SLOPE STABILITY BASED ON GA-BP HYBRID ALGORITHM.
Chen et al. Data-driven distribution network topology identification considering correlated generation power of distributed energy resource
CN116316617B (en) Multi-station intelligent fusion new energy generation power region prediction method and system
Chen et al. Parameter Sensitivity and Inversion Analysis for a Concrete Face Rockfill Dam Based on CS‐BPNN
CN110852610A (en) Road, bridge and tunnel health state and maintenance cost measuring and calculating method based on Markov model
Bontempi et al. Semiphysical modelling architecture for dynamic assessment of power components loading capability
CN115217152A (en) Method and device for predicting opening and closing deformation of immersed tunnel pipe joint
CN112926801B (en) Load curve combined prediction method and device based on quantile regression
CN109213800A (en) A kind of tobacco insect pest situation forecasting system and method
Li et al. Hydropower generation forecasting via deep neural network
Hu et al. Research on ACO-BP based prediction method of the oilfield production stimulation results
CN105117605B (en) A kind of device and method of case prediction
CN113673145A (en) Surface subsidence prediction method based on PCA-LSTM neural network
CN110556595B (en) Method for predicting quality unbalance of lithium iron phosphate battery safety management system for energy storage
CN114036723A (en) Method, device, equipment and storage medium for predicting running state of comprehensive energy system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20200609